French g2p with pronunciation dictionary

[mgrafu]

What does this PR do ?

Add a one line overview of what this PR aims to accomplish.

Collection: [Note which collection this PR will affect]

Changelog

• Add specific line by line info of high level changes in this PR.

Usage

• You can potentially add a usage example below

# Add a code snippet demonstrating how to use this 

Before your PR is "Ready for review"

Pre checks:

• Make sure you read and followed Contributor guidelines
• Did you write any new necessary tests?
• Did you add or update any necessary documentation?
• Does the PR affect components that are optional to install? (Ex: Numba, Pynini, Apex etc)
  • Reviewer: Does the PR have correct import guards for all optional libraries?

PR Type:

• New Feature
• Bugfix
• Documentation

If you haven't finished some of the above items you can still open "Draft" PR.

Who can review?

Anyone in the community is free to review the PR once the checks have passed.
Contributor guidelines contains specific people who can review PRs to various areas.

Additional Information

• Related to # (issue)

[XuesongYang]

LGTM. Could you please also add some unit tests for the tokenizers?

• https://github.com/NVIDIA/NeMo/blob/main/tests/collections/common/tokenizers/text_to_speech/test_tokenizer_utils.py
• https://github.com/NVIDIA/NeMo/blob/main/tests/collections/common/tokenizers/text_to_speech/test_tts_tokenizers.py

[XuesongYang]

these tests are for text_preprocessing funcs, not for any tokenize funcs. Could you pls revise following the above unit tests?

[XuesongYang]

we can remove the above three unit tests, and instead add test examples inside all functions related to any_locale_word_tokenize. For example, you could extend below input_text and expected_output as a list by adding fr-fr examples.

@pytest.mark.run_only_on('CPU')
    @pytest.mark.unit
    def test_any_locale_word_tokenize(self):
        input_text = "apple banana pear"
        expected_output = self._create_expected_output(["apple", " ", "banana", " ", "pear"])

        output = any_locale_word_tokenize(input_text)
        assert output == expected_output

    @pytest.mark.run_only_on('CPU')
    @pytest.mark.unit
    def test_any_locale_word_tokenize_with_accents(self):
        input_text = "The naïve piñata at the café..."
        expected_output = self._create_expected_output(
            ["The", " ", "naïve", " ", "piñata", " ", "at", " ", "the", " ", "café", "..."]
        )

        output = any_locale_word_tokenize(input_text)
        assert output == expected_output

    @pytest.mark.run_only_on('CPU')
    @pytest.mark.unit
    def test_any_locale_word_tokenize_with_numbers(self):
        input_text = r"Three times× four^teen ÷divided by [movies] on \slash."
        expected_output = self._create_expected_output(
            [
                "Three",
                " ",
                "times",
                "× ",
                "four",
                "^",
                "teen",
                " ÷",
                "divided",
                " ",
                "by",
                " [",
                "movies",
                "] ",
                "on",
                " \\",
                "slash",
                ".",
            ]
        )

        output = any_locale_word_tokenize(input_text)
        assert output == expected_output

[XuesongYang]

when you specified case="mixed", meaning the tokenizer will distinguish uppercase and lowercase letters, and assign different token index. For example, we should expect the final char alphabet like A-Za-z. But from your test cases for French, I feel all are lowercase. Could you please double check either you want all lowercased, uppercased, or mixed?

[mgrafu]

corrected this in newer commit :)

[XuesongYang]

Regarding the extra punctuations used in French, could you pls add comments of unicode for each punctuation just as what "de-DE" did? It is not easy to pinpoint the difference between similar surfaces, such as below. Thanks!

'‒',  # figure dash, U+2012, decimal 8210
'–',  # en dash, U+2013, decimal 8211

[XuesongYang]

LGTM. Thanks!